tcp_usrreq.c revision 1.166.4.3
1 /* $NetBSD: tcp_usrreq.c,v 1.166.4.3 2013/09/23 00:57:53 rmind Exp $ */ 2 3 /* 4 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 3. Neither the name of the project nor the names of its contributors 16 * may be used to endorse or promote products derived from this software 17 * without specific prior written permission. 18 * 19 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND 20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 22 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE 23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 29 * SUCH DAMAGE. 30 */ 31 32 /*- 33 * Copyright (c) 1997, 1998, 2005, 2006 The NetBSD Foundation, Inc. 34 * All rights reserved. 35 * 36 * This code is derived from software contributed to The NetBSD Foundation 37 * by Jason R. Thorpe and Kevin M. Lahey of the Numerical Aerospace Simulation 38 * Facility, NASA Ames Research Center. 39 * This code is derived from software contributed to The NetBSD Foundation 40 * by Charles M. Hannum. 41 * This code is derived from software contributed to The NetBSD Foundation 42 * by Rui Paulo. 43 * 44 * Redistribution and use in source and binary forms, with or without 45 * modification, are permitted provided that the following conditions 46 * are met: 47 * 1. Redistributions of source code must retain the above copyright 48 * notice, this list of conditions and the following disclaimer. 49 * 2. Redistributions in binary form must reproduce the above copyright 50 * notice, this list of conditions and the following disclaimer in the 51 * documentation and/or other materials provided with the distribution. 52 * 53 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 54 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 55 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 56 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 57 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 58 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 59 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 60 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 61 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 62 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 63 * POSSIBILITY OF SUCH DAMAGE. 64 */ 65 66 /* 67 * Copyright (c) 1982, 1986, 1988, 1993, 1995 68 * The Regents of the University of California. All rights reserved. 69 * 70 * Redistribution and use in source and binary forms, with or without 71 * modification, are permitted provided that the following conditions 72 * are met: 73 * 1. Redistributions of source code must retain the above copyright 74 * notice, this list of conditions and the following disclaimer. 75 * 2. Redistributions in binary form must reproduce the above copyright 76 * notice, this list of conditions and the following disclaimer in the 77 * documentation and/or other materials provided with the distribution. 78 * 3. Neither the name of the University nor the names of its contributors 79 * may be used to endorse or promote products derived from this software 80 * without specific prior written permission. 81 * 82 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 83 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 84 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 85 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 86 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 87 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 88 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 89 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 90 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 91 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 92 * SUCH DAMAGE. 93 * 94 * @(#)tcp_usrreq.c 8.5 (Berkeley) 6/21/95 95 */ 96 97 #include <sys/cdefs.h> 98 __KERNEL_RCSID(0, "$NetBSD: tcp_usrreq.c,v 1.166.4.3 2013/09/23 00:57:53 rmind Exp $"); 99 100 #include "opt_inet.h" 101 #include "opt_ipsec.h" 102 #include "opt_tcp_debug.h" 103 #include "opt_mbuftrace.h" 104 105 106 #include <sys/param.h> 107 #include <sys/systm.h> 108 #include <sys/kernel.h> 109 #include <sys/malloc.h> 110 #include <sys/mbuf.h> 111 #include <sys/socket.h> 112 #include <sys/socketvar.h> 113 #include <sys/protosw.h> 114 #include <sys/errno.h> 115 #include <sys/stat.h> 116 #include <sys/proc.h> 117 #include <sys/domain.h> 118 #include <sys/sysctl.h> 119 #include <sys/kauth.h> 120 #include <sys/uidinfo.h> 121 122 #include <net/if.h> 123 #include <net/route.h> 124 125 #include <netinet/in.h> 126 #include <netinet/in_systm.h> 127 #include <netinet/in_var.h> 128 #include <netinet/ip.h> 129 #include <netinet/in_pcb.h> 130 #include <netinet/ip_var.h> 131 #include <netinet/in_offload.h> 132 133 #ifdef INET6 134 #ifndef INET 135 #include <netinet/in.h> 136 #endif 137 #include <netinet/ip6.h> 138 #include <netinet6/in6_pcb.h> 139 #include <netinet6/ip6_var.h> 140 #include <netinet6/scope6_var.h> 141 #endif 142 143 #include <netinet/tcp.h> 144 #include <netinet/tcp_fsm.h> 145 #include <netinet/tcp_seq.h> 146 #include <netinet/tcp_timer.h> 147 #include <netinet/tcp_var.h> 148 #include <netinet/tcp_private.h> 149 #include <netinet/tcp_congctl.h> 150 #include <netinet/tcpip.h> 151 #include <netinet/tcp_debug.h> 152 #include <netinet/tcp_vtw.h> 153 154 #include "opt_tcp_space.h" 155 156 /* 157 * TCP protocol interface to socket abstraction. 158 */ 159 160 /* 161 * Process a TCP user request for TCP tb. If this is a send request 162 * then m is the mbuf chain of send data. If this is a timer expiration 163 * (called from the software clock routine), then timertype tells which timer. 164 */ 165 static int 166 tcp_usrreq(struct socket *so, int req, struct mbuf *m, struct mbuf *nam, 167 struct mbuf *control, struct lwp *l) 168 { 169 inpcb_t *inp; 170 #ifdef INET6 171 struct in6pcb *in6p; 172 #endif 173 struct tcpcb *tp = NULL; 174 int s; 175 int error = 0; 176 #ifdef TCP_DEBUG 177 int ostate = 0; 178 #endif 179 int family; /* family of the socket */ 180 181 KASSERT(req != PRU_ATTACH); 182 KASSERT(req != PRU_DETACH); 183 184 family = so->so_proto->pr_domain->dom_family; 185 186 if (req == PRU_CONTROL) { 187 switch (family) { 188 #ifdef INET 189 case PF_INET: 190 return (in_control(so, (long)m, (void *)nam, 191 (struct ifnet *)control, l)); 192 #endif 193 #ifdef INET6 194 case PF_INET6: 195 return (in6_control(so, (long)m, (void *)nam, 196 (struct ifnet *)control, l)); 197 #endif 198 default: 199 return EAFNOSUPPORT; 200 } 201 } 202 203 s = splsoftnet(); 204 205 if (req == PRU_PURGEIF) { 206 mutex_enter(softnet_lock); 207 switch (family) { 208 #ifdef INET 209 case PF_INET: 210 inpcb_purgeif0(tcbtable, (struct ifnet *)control); 211 in_purgeif((struct ifnet *)control); 212 inpcb_purgeif(tcbtable, (struct ifnet *)control); 213 break; 214 #endif 215 #ifdef INET6 216 case PF_INET6: 217 in6_pcbpurgeif0(tcbtable, (struct ifnet *)control); 218 in6_purgeif((struct ifnet *)control); 219 in6_pcbpurgeif(tcbtable, (struct ifnet *)control); 220 break; 221 #endif 222 default: 223 mutex_exit(softnet_lock); 224 splx(s); 225 return (EAFNOSUPPORT); 226 } 227 mutex_exit(softnet_lock); 228 splx(s); 229 return (0); 230 } 231 232 KASSERT(solocked(so)); 233 234 switch (family) { 235 #ifdef INET 236 case PF_INET: 237 inp = sotoinpcb(so); 238 #ifdef INET6 239 in6p = NULL; 240 #endif 241 break; 242 #endif 243 #ifdef INET6 244 case PF_INET6: 245 inp = NULL; 246 in6p = sotoin6pcb(so); 247 break; 248 #endif 249 default: 250 splx(s); 251 return EAFNOSUPPORT; 252 } 253 254 #ifdef INET6 255 KASSERT((inp != NULL) ^ (in6p != NULL)); 256 #endif 257 KASSERT(!control || (req == PRU_SEND || req == PRU_SENDOOB)); 258 259 /* 260 * When a TCP is attached to a socket, then there will be 261 * a (struct inpcb) pointed at by the socket, and this 262 * structure will point at a subsidary (struct tcpcb). 263 */ 264 if ((inp == NULL 265 #ifdef INET6 266 && in6p == NULL 267 #endif 268 ) && req != PRU_SENSE) 269 { 270 error = EINVAL; 271 goto release; 272 } 273 #ifdef INET 274 if (inp) { 275 tp = intotcpcb(inp); 276 /* WHAT IF TP IS 0? */ 277 #ifdef KPROF 278 tcp_acounts[tp->t_state][req]++; 279 #endif 280 #ifdef TCP_DEBUG 281 ostate = tp->t_state; 282 #endif 283 } 284 #endif 285 #ifdef INET6 286 if (in6p) { 287 tp = in6totcpcb(in6p); 288 /* WHAT IF TP IS 0? */ 289 #ifdef KPROF 290 tcp_acounts[tp->t_state][req]++; 291 #endif 292 #ifdef TCP_DEBUG 293 ostate = tp->t_state; 294 #endif 295 } 296 #endif 297 298 switch (req) { 299 /* 300 * Give the socket an address. 301 */ 302 case PRU_BIND: 303 switch (family) { 304 #ifdef INET 305 case PF_INET: 306 error = inpcb_bind(inp, nam, l); 307 break; 308 #endif 309 #ifdef INET6 310 case PF_INET6: 311 error = in6_pcbbind(in6p, nam, l); 312 if (!error) { 313 /* mapped addr case */ 314 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr)) 315 tp->t_family = AF_INET; 316 else 317 tp->t_family = AF_INET6; 318 } 319 break; 320 #endif 321 } 322 break; 323 324 /* 325 * Prepare to accept connections. 326 */ 327 case PRU_LISTEN: 328 #ifdef INET 329 if (inp) { 330 in_port_t lport; 331 332 inpcb_get_ports(inp, &lport, NULL); 333 if (lport == 0) { 334 error = inpcb_bind(inp, NULL, l); 335 if (error) 336 break; 337 } 338 } 339 #endif 340 #ifdef INET6 341 if (in6p && in6p->in6p_lport == 0) { 342 error = in6_pcbbind(in6p, NULL, l); 343 if (error) 344 break; 345 } 346 #endif 347 tp->t_state = TCPS_LISTEN; 348 break; 349 350 /* 351 * Initiate connection to peer. 352 * Create a template for use in transmissions on this connection. 353 * Enter SYN_SENT state, and mark socket as connecting. 354 * Start keep-alive timer, and seed output sequence space. 355 * Send initial segment on connection. 356 */ 357 case PRU_CONNECT: 358 #ifdef INET 359 if (inp) { 360 in_port_t lport; 361 362 inpcb_get_ports(inp, &lport, NULL); 363 if (lport == 0) { 364 error = inpcb_bind(inp, NULL, l); 365 if (error) 366 break; 367 } 368 error = inpcb_connect(inp, nam, l); 369 } 370 #endif 371 #ifdef INET6 372 if (in6p) { 373 if (in6p->in6p_lport == 0) { 374 error = in6_pcbbind(in6p, NULL, l); 375 if (error) 376 break; 377 } 378 error = in6_pcbconnect(in6p, nam, l); 379 if (!error) { 380 /* mapped addr case */ 381 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr)) 382 tp->t_family = AF_INET; 383 else 384 tp->t_family = AF_INET6; 385 } 386 } 387 #endif 388 if (error) 389 break; 390 tp->t_template = tcp_template(tp); 391 if (tp->t_template == 0) { 392 #ifdef INET 393 if (inp) 394 inpcb_disconnect(inp); 395 #endif 396 #ifdef INET6 397 if (in6p) 398 in6_pcbdisconnect(in6p); 399 #endif 400 error = ENOBUFS; 401 break; 402 } 403 /* 404 * Compute window scaling to request. 405 * XXX: This should be moved to tcp_output(). 406 */ 407 while (tp->request_r_scale < TCP_MAX_WINSHIFT && 408 (TCP_MAXWIN << tp->request_r_scale) < sb_max) 409 tp->request_r_scale++; 410 soisconnecting(so); 411 TCP_STATINC(TCP_STAT_CONNATTEMPT); 412 tp->t_state = TCPS_SYN_SENT; 413 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepinit); 414 tp->iss = tcp_new_iss(tp, 0); 415 tcp_sendseqinit(tp); 416 error = tcp_output(tp); 417 break; 418 419 /* 420 * Create a TCP connection between two sockets. 421 */ 422 case PRU_CONNECT2: 423 error = EOPNOTSUPP; 424 break; 425 426 /* 427 * Initiate disconnect from peer. 428 * If connection never passed embryonic stage, just drop; 429 * else if don't need to let data drain, then can just drop anyways, 430 * else have to begin TCP shutdown process: mark socket disconnecting, 431 * drain unread data, state switch to reflect user close, and 432 * send segment (e.g. FIN) to peer. Socket will be really disconnected 433 * when peer sends FIN and acks ours. 434 * 435 * SHOULD IMPLEMENT LATER PRU_CONNECT VIA REALLOC TCPCB. 436 */ 437 case PRU_DISCONNECT: 438 tp = tcp_disconnect(tp); 439 break; 440 441 /* 442 * Accept a connection. Essentially all the work is 443 * done at higher levels; just return the address 444 * of the peer, storing through addr. 445 */ 446 case PRU_ACCEPT: 447 #ifdef INET 448 if (inp) 449 inpcb_fetch_peeraddr(inp, nam); 450 #endif 451 #ifdef INET6 452 if (in6p) 453 in6_setpeeraddr(in6p, nam); 454 #endif 455 break; 456 457 /* 458 * Mark the connection as being incapable of further output. 459 */ 460 case PRU_SHUTDOWN: 461 socantsendmore(so); 462 tp = tcp_usrclosed(tp); 463 if (tp) 464 error = tcp_output(tp); 465 break; 466 467 /* 468 * After a receive, possibly send window update to peer. 469 */ 470 case PRU_RCVD: 471 /* 472 * soreceive() calls this function when a user receives 473 * ancillary data on a listening socket. We don't call 474 * tcp_output in such a case, since there is no header 475 * template for a listening socket and hence the kernel 476 * will panic. 477 */ 478 if ((so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING)) != 0) 479 (void) tcp_output(tp); 480 break; 481 482 /* 483 * Do a send by putting data in output queue and updating urgent 484 * marker if URG set. Possibly send more data. 485 */ 486 case PRU_SEND: 487 if (control && control->m_len) { 488 m_freem(control); 489 m_freem(m); 490 error = EINVAL; 491 break; 492 } 493 sbappendstream(&so->so_snd, m); 494 error = tcp_output(tp); 495 break; 496 497 /* 498 * Abort the TCP. 499 */ 500 case PRU_ABORT: 501 tp = tcp_drop(tp, ECONNABORTED); 502 break; 503 504 case PRU_SENSE: 505 /* 506 * stat: don't bother with a blocksize. 507 */ 508 splx(s); 509 return (0); 510 511 case PRU_RCVOOB: 512 if (control && control->m_len) { 513 m_freem(control); 514 m_freem(m); 515 error = EINVAL; 516 break; 517 } 518 if ((so->so_oobmark == 0 && 519 (so->so_state & SS_RCVATMARK) == 0) || 520 so->so_options & SO_OOBINLINE || 521 tp->t_oobflags & TCPOOB_HADDATA) { 522 error = EINVAL; 523 break; 524 } 525 if ((tp->t_oobflags & TCPOOB_HAVEDATA) == 0) { 526 error = EWOULDBLOCK; 527 break; 528 } 529 m->m_len = 1; 530 *mtod(m, char *) = tp->t_iobc; 531 if (((long)nam & MSG_PEEK) == 0) 532 tp->t_oobflags ^= (TCPOOB_HAVEDATA | TCPOOB_HADDATA); 533 break; 534 535 case PRU_SENDOOB: 536 if (sbspace(&so->so_snd) < -512) { 537 m_freem(m); 538 error = ENOBUFS; 539 break; 540 } 541 /* 542 * According to RFC961 (Assigned Protocols), 543 * the urgent pointer points to the last octet 544 * of urgent data. We continue, however, 545 * to consider it to indicate the first octet 546 * of data past the urgent section. 547 * Otherwise, snd_up should be one lower. 548 */ 549 sbappendstream(&so->so_snd, m); 550 tp->snd_up = tp->snd_una + so->so_snd.sb_cc; 551 tp->t_force = 1; 552 error = tcp_output(tp); 553 tp->t_force = 0; 554 break; 555 556 case PRU_SOCKADDR: 557 #ifdef INET 558 if (inp) 559 inpcb_fetch_sockaddr(inp, nam); 560 #endif 561 #ifdef INET6 562 if (in6p) 563 in6_setsockaddr(in6p, nam); 564 #endif 565 break; 566 567 case PRU_PEERADDR: 568 #ifdef INET 569 if (inp) 570 inpcb_fetch_peeraddr(inp, nam); 571 #endif 572 #ifdef INET6 573 if (in6p) 574 in6_setpeeraddr(in6p, nam); 575 #endif 576 break; 577 578 default: 579 panic("tcp_usrreq"); 580 } 581 #ifdef TCP_DEBUG 582 if (tp && (so->so_options & SO_DEBUG)) 583 tcp_trace(TA_USER, ostate, tp, NULL, req); 584 #endif 585 586 release: 587 splx(s); 588 return (error); 589 } 590 591 static void 592 change_keepalive(struct socket *so, struct tcpcb *tp) 593 { 594 tp->t_maxidle = tp->t_keepcnt * tp->t_keepintvl; 595 TCP_TIMER_DISARM(tp, TCPT_KEEP); 596 TCP_TIMER_DISARM(tp, TCPT_2MSL); 597 598 if (tp->t_state == TCPS_SYN_RECEIVED || 599 tp->t_state == TCPS_SYN_SENT) { 600 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepinit); 601 } else if (so->so_options & SO_KEEPALIVE && 602 tp->t_state <= TCPS_CLOSE_WAIT) { 603 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepintvl); 604 } else { 605 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepidle); 606 } 607 608 if ((tp->t_state == TCPS_FIN_WAIT_2) && (tp->t_maxidle > 0)) 609 TCP_TIMER_ARM(tp, TCPT_2MSL, tp->t_maxidle); 610 } 611 612 613 int 614 tcp_ctloutput(int op, struct socket *so, struct sockopt *sopt) 615 { 616 int error = 0, s; 617 inpcb_t *inp; 618 #ifdef INET6 619 struct in6pcb *in6p; 620 #endif 621 struct tcpcb *tp; 622 u_int ui; 623 int family; /* family of the socket */ 624 int level, optname, optval; 625 626 level = sopt->sopt_level; 627 optname = sopt->sopt_name; 628 629 family = so->so_proto->pr_domain->dom_family; 630 631 s = splsoftnet(); 632 switch (family) { 633 #ifdef INET 634 case PF_INET: 635 inp = sotoinpcb(so); 636 #ifdef INET6 637 in6p = NULL; 638 #endif 639 break; 640 #endif 641 #ifdef INET6 642 case PF_INET6: 643 inp = NULL; 644 in6p = sotoin6pcb(so); 645 break; 646 #endif 647 default: 648 splx(s); 649 panic("%s: af %d", __func__, family); 650 } 651 #ifndef INET6 652 if (inp == NULL) 653 #else 654 if (inp == NULL && in6p == NULL) 655 #endif 656 { 657 splx(s); 658 return (ECONNRESET); 659 } 660 if (level != IPPROTO_TCP) { 661 switch (family) { 662 #ifdef INET 663 case PF_INET: 664 error = ip_ctloutput(op, so, sopt); 665 break; 666 #endif 667 #ifdef INET6 668 case PF_INET6: 669 error = ip6_ctloutput(op, so, sopt); 670 break; 671 #endif 672 } 673 splx(s); 674 return (error); 675 } 676 if (inp) 677 tp = intotcpcb(inp); 678 #ifdef INET6 679 else if (in6p) 680 tp = in6totcpcb(in6p); 681 #endif 682 else 683 tp = NULL; 684 685 switch (op) { 686 case PRCO_SETOPT: 687 switch (optname) { 688 #ifdef TCP_SIGNATURE 689 case TCP_MD5SIG: 690 error = sockopt_getint(sopt, &optval); 691 if (error) 692 break; 693 if (optval > 0) 694 tp->t_flags |= TF_SIGNATURE; 695 else 696 tp->t_flags &= ~TF_SIGNATURE; 697 break; 698 #endif /* TCP_SIGNATURE */ 699 700 case TCP_NODELAY: 701 error = sockopt_getint(sopt, &optval); 702 if (error) 703 break; 704 if (optval) 705 tp->t_flags |= TF_NODELAY; 706 else 707 tp->t_flags &= ~TF_NODELAY; 708 break; 709 710 case TCP_MAXSEG: 711 error = sockopt_getint(sopt, &optval); 712 if (error) 713 break; 714 if (optval > 0 && optval <= tp->t_peermss) 715 tp->t_peermss = optval; /* limit on send size */ 716 else 717 error = EINVAL; 718 break; 719 #ifdef notyet 720 case TCP_CONGCTL: 721 /* XXX string overflow XXX */ 722 error = tcp_congctl_select(tp, sopt->sopt_data); 723 break; 724 #endif 725 726 case TCP_KEEPIDLE: 727 error = sockopt_get(sopt, &ui, sizeof(ui)); 728 if (error) 729 break; 730 if (ui > 0) { 731 tp->t_keepidle = ui; 732 change_keepalive(so, tp); 733 } else 734 error = EINVAL; 735 break; 736 737 case TCP_KEEPINTVL: 738 error = sockopt_get(sopt, &ui, sizeof(ui)); 739 if (error) 740 break; 741 if (ui > 0) { 742 tp->t_keepintvl = ui; 743 change_keepalive(so, tp); 744 } else 745 error = EINVAL; 746 break; 747 748 case TCP_KEEPCNT: 749 error = sockopt_get(sopt, &ui, sizeof(ui)); 750 if (error) 751 break; 752 if (ui > 0) { 753 tp->t_keepcnt = ui; 754 change_keepalive(so, tp); 755 } else 756 error = EINVAL; 757 break; 758 759 case TCP_KEEPINIT: 760 error = sockopt_get(sopt, &ui, sizeof(ui)); 761 if (error) 762 break; 763 if (ui > 0) { 764 tp->t_keepinit = ui; 765 change_keepalive(so, tp); 766 } else 767 error = EINVAL; 768 break; 769 770 default: 771 error = ENOPROTOOPT; 772 break; 773 } 774 break; 775 776 case PRCO_GETOPT: 777 switch (optname) { 778 #ifdef TCP_SIGNATURE 779 case TCP_MD5SIG: 780 optval = (tp->t_flags & TF_SIGNATURE) ? 1 : 0; 781 error = sockopt_set(sopt, &optval, sizeof(optval)); 782 break; 783 #endif 784 case TCP_NODELAY: 785 optval = tp->t_flags & TF_NODELAY; 786 error = sockopt_set(sopt, &optval, sizeof(optval)); 787 break; 788 case TCP_MAXSEG: 789 optval = tp->t_peermss; 790 error = sockopt_set(sopt, &optval, sizeof(optval)); 791 break; 792 #ifdef notyet 793 case TCP_CONGCTL: 794 break; 795 #endif 796 default: 797 error = ENOPROTOOPT; 798 break; 799 } 800 break; 801 } 802 splx(s); 803 return (error); 804 } 805 806 #ifndef TCP_SENDSPACE 807 #define TCP_SENDSPACE 1024*32 808 #endif 809 int tcp_sendspace = TCP_SENDSPACE; 810 #ifndef TCP_RECVSPACE 811 #define TCP_RECVSPACE 1024*32 812 #endif 813 int tcp_recvspace = TCP_RECVSPACE; 814 815 /* 816 * tcp_attach: attach TCP protocol to socket, allocating internet protocol 817 * control block, TCP control block, buffer space and entering LISTEN state 818 * if to accept connections. 819 */ 820 static int 821 tcp_attach(struct socket *so, int proto) 822 { 823 inpcb_t *inp; 824 struct tcpcb *tp; 825 #ifdef INET6 826 struct in6pcb *in6p; 827 #endif 828 int s, error, family; 829 830 s = splsoftnet(); 831 sosetlock(so); 832 solock(so); 833 834 family = so->so_proto->pr_domain->dom_family; 835 switch (family) { 836 #ifdef INET 837 case PF_INET: 838 inp = sotoinpcb(so); 839 #ifdef INET6 840 in6p = NULL; 841 #endif 842 break; 843 #endif 844 #ifdef INET6 845 case PF_INET6: 846 inp = NULL; 847 in6p = sotoin6pcb(so); 848 break; 849 #endif 850 default: 851 error = EAFNOSUPPORT; 852 goto out; 853 } 854 855 KASSERT(inp == NULL); 856 #ifdef INET6 857 KASSERT(in6p == NULL); 858 #endif 859 860 #ifdef MBUFTRACE 861 so->so_mowner = &tcp_sock_mowner; 862 so->so_rcv.sb_mowner = &tcp_sock_rx_mowner; 863 so->so_snd.sb_mowner = &tcp_sock_tx_mowner; 864 #endif 865 if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) { 866 error = soreserve(so, tcp_sendspace, tcp_recvspace); 867 if (error) { 868 goto out; 869 } 870 } 871 872 so->so_rcv.sb_flags |= SB_AUTOSIZE; 873 so->so_snd.sb_flags |= SB_AUTOSIZE; 874 875 tp = NULL; 876 switch (family) { 877 #ifdef INET 878 case PF_INET: 879 error = inpcb_create(so, tcbtable); 880 if ((inp = sotoinpcb(so)) != NULL) { 881 tp = tcp_newtcpcb(family, (void *)inp); 882 } 883 #ifdef INET6 884 in6p = NULL; 885 #endif 886 break; 887 #endif 888 #ifdef INET6 889 case PF_INET6: 890 error = in6_pcballoc(so, tcbtable); 891 if ((in6p = sotoin6pcb(so)) != NULL) { 892 tp = tcp_newtcpcb(family, (void *)in6p); 893 } 894 inp = NULL; 895 break; 896 #endif 897 default: 898 error = EAFNOSUPPORT; 899 } 900 901 if (error) { 902 goto out; 903 } 904 905 if (tp == NULL) { 906 int nofd = so->so_state & SS_NOFDREF; /* XXX */ 907 908 so->so_state &= ~SS_NOFDREF; /* don't free the socket yet */ 909 #ifdef INET 910 if (inp) 911 inpcb_destroy(inp); 912 #endif 913 #ifdef INET6 914 if (in6p) 915 in6_pcbdetach(in6p); 916 #endif 917 so->so_state |= nofd; 918 error = ENOBUFS; 919 goto out; 920 } 921 tp->t_state = TCPS_CLOSED; 922 923 if ((so->so_options & SO_LINGER) && so->so_linger == 0) 924 so->so_linger = TCP_LINGERTIME; 925 tp = sototcpcb(so); 926 out: 927 sounlock(so); 928 splx(s); 929 return error; 930 } 931 932 static void 933 tcp_detach(struct socket *so) 934 { 935 inpcb_t *inp; 936 #ifdef INET6 937 struct in6pcb *in6p; 938 #endif 939 struct tcpcb *tp = NULL; 940 int s, family; 941 942 KASSERT(solocked(so)); 943 944 s = splsoftnet(); 945 family = so->so_proto->pr_domain->dom_family; 946 switch (family) { 947 #ifdef INET 948 case PF_INET: 949 inp = sotoinpcb(so); 950 tp = intotcpcb(inp); 951 break; 952 #endif 953 #ifdef INET6 954 case PF_INET6: 955 in6p = sotoin6pcb(so); 956 tp = in6totcpcb(in6p); 957 break; 958 #endif 959 default: 960 splx(s); 961 return; 962 } 963 KASSERT(tp != NULL); 964 (void)tcp_disconnect(tp); 965 splx(s); 966 } 967 968 /* 969 * Initiate (or continue) disconnect. 970 * If embryonic state, just send reset (once). 971 * If in ``let data drain'' option and linger null, just drop. 972 * Otherwise (hard), mark socket disconnecting and drop 973 * current input data; switch states based on user close, and 974 * send segment to peer (with FIN). 975 */ 976 struct tcpcb * 977 tcp_disconnect(struct tcpcb *tp) 978 { 979 struct socket *so; 980 981 if (tp->t_inpcb) 982 so = inpcb_get_socket(tp->t_inpcb); 983 #ifdef INET6 984 else if (tp->t_in6pcb) 985 so = tp->t_in6pcb->in6p_socket; 986 #endif 987 else 988 so = NULL; 989 990 KASSERT(so != NULL); 991 992 if (TCPS_HAVEESTABLISHED(tp->t_state) == 0) 993 tp = tcp_close(tp); 994 else if ((so->so_options & SO_LINGER) && so->so_linger == 0) 995 tp = tcp_drop(tp, 0); 996 else { 997 soisdisconnecting(so); 998 sbflush(&so->so_rcv); 999 tp = tcp_usrclosed(tp); 1000 if (tp) 1001 (void) tcp_output(tp); 1002 } 1003 return (tp); 1004 } 1005 1006 /* 1007 * User issued close, and wish to trail through shutdown states: 1008 * if never received SYN, just forget it. If got a SYN from peer, 1009 * but haven't sent FIN, then go to FIN_WAIT_1 state to send peer a FIN. 1010 * If already got a FIN from peer, then almost done; go to LAST_ACK 1011 * state. In all other cases, have already sent FIN to peer (e.g. 1012 * after PRU_SHUTDOWN), and just have to play tedious game waiting 1013 * for peer to send FIN or not respond to keep-alives, etc. 1014 * We can let the user exit from the close as soon as the FIN is acked. 1015 */ 1016 struct tcpcb * 1017 tcp_usrclosed(struct tcpcb *tp) 1018 { 1019 1020 switch (tp->t_state) { 1021 1022 case TCPS_CLOSED: 1023 case TCPS_LISTEN: 1024 case TCPS_SYN_SENT: 1025 tp->t_state = TCPS_CLOSED; 1026 tp = tcp_close(tp); 1027 break; 1028 1029 case TCPS_SYN_RECEIVED: 1030 case TCPS_ESTABLISHED: 1031 tp->t_state = TCPS_FIN_WAIT_1; 1032 break; 1033 1034 case TCPS_CLOSE_WAIT: 1035 tp->t_state = TCPS_LAST_ACK; 1036 break; 1037 } 1038 if (tp && tp->t_state >= TCPS_FIN_WAIT_2) { 1039 struct socket *so; 1040 if (tp->t_inpcb) 1041 so = inpcb_get_socket(tp->t_inpcb); 1042 #ifdef INET6 1043 else if (tp->t_in6pcb) 1044 so = tp->t_in6pcb->in6p_socket; 1045 #endif 1046 else 1047 so = NULL; 1048 if (so) 1049 soisdisconnected(so); 1050 /* 1051 * If we are in FIN_WAIT_2, we arrived here because the 1052 * application did a shutdown of the send side. Like the 1053 * case of a transition from FIN_WAIT_1 to FIN_WAIT_2 after 1054 * a full close, we start a timer to make sure sockets are 1055 * not left in FIN_WAIT_2 forever. 1056 */ 1057 if ((tp->t_state == TCPS_FIN_WAIT_2) && (tp->t_maxidle > 0)) 1058 TCP_TIMER_ARM(tp, TCPT_2MSL, tp->t_maxidle); 1059 else if (tp->t_state == TCPS_TIME_WAIT 1060 && ((tp->t_inpcb 1061 && (tcp4_vtw_enable & 1) 1062 && vtw_add(AF_INET, tp)) 1063 || 1064 (tp->t_in6pcb 1065 && (tcp6_vtw_enable & 1) 1066 && vtw_add(AF_INET6, tp)))) { 1067 tp = 0; 1068 } 1069 } 1070 return (tp); 1071 } 1072 1073 /* 1074 * sysctl helper routine for net.inet.ip.mssdflt. it can't be less 1075 * than 32. 1076 */ 1077 static int 1078 sysctl_net_inet_tcp_mssdflt(SYSCTLFN_ARGS) 1079 { 1080 int error, mssdflt; 1081 struct sysctlnode node; 1082 1083 mssdflt = tcp_mssdflt; 1084 node = *rnode; 1085 node.sysctl_data = &mssdflt; 1086 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1087 if (error || newp == NULL) 1088 return (error); 1089 1090 if (mssdflt < 32) 1091 return (EINVAL); 1092 tcp_mssdflt = mssdflt; 1093 1094 return (0); 1095 } 1096 1097 /* 1098 * sysctl helper routine for setting port related values under 1099 * net.inet.ip and net.inet6.ip6. does basic range checking and does 1100 * additional checks for each type. this code has placed in 1101 * tcp_input.c since INET and INET6 both use the same tcp code. 1102 * 1103 * this helper is not static so that both inet and inet6 can use it. 1104 */ 1105 int 1106 sysctl_net_inet_ip_ports(SYSCTLFN_ARGS) 1107 { 1108 int error, tmp; 1109 int apmin, apmax; 1110 #ifndef IPNOPRIVPORTS 1111 int lpmin, lpmax; 1112 #endif /* IPNOPRIVPORTS */ 1113 struct sysctlnode node; 1114 1115 if (namelen != 0) 1116 return (EINVAL); 1117 1118 switch (name[-3]) { 1119 #ifdef INET 1120 case PF_INET: 1121 apmin = anonportmin; 1122 apmax = anonportmax; 1123 #ifndef IPNOPRIVPORTS 1124 lpmin = lowportmin; 1125 lpmax = lowportmax; 1126 #endif /* IPNOPRIVPORTS */ 1127 break; 1128 #endif /* INET */ 1129 #ifdef INET6 1130 case PF_INET6: 1131 apmin = ip6_anonportmin; 1132 apmax = ip6_anonportmax; 1133 #ifndef IPNOPRIVPORTS 1134 lpmin = ip6_lowportmin; 1135 lpmax = ip6_lowportmax; 1136 #endif /* IPNOPRIVPORTS */ 1137 break; 1138 #endif /* INET6 */ 1139 default: 1140 return (EINVAL); 1141 } 1142 1143 /* 1144 * insert temporary copy into node, perform lookup on 1145 * temporary, then restore pointer 1146 */ 1147 node = *rnode; 1148 tmp = *(int*)rnode->sysctl_data; 1149 node.sysctl_data = &tmp; 1150 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1151 if (error || newp == NULL) 1152 return (error); 1153 1154 /* 1155 * simple port range check 1156 */ 1157 if (tmp < 0 || tmp > 65535) 1158 return (EINVAL); 1159 1160 /* 1161 * per-node range checks 1162 */ 1163 switch (rnode->sysctl_num) { 1164 case IPCTL_ANONPORTMIN: 1165 case IPV6CTL_ANONPORTMIN: 1166 if (tmp >= apmax) 1167 return (EINVAL); 1168 #ifndef IPNOPRIVPORTS 1169 if (tmp < IPPORT_RESERVED) 1170 return (EINVAL); 1171 #endif /* IPNOPRIVPORTS */ 1172 break; 1173 1174 case IPCTL_ANONPORTMAX: 1175 case IPV6CTL_ANONPORTMAX: 1176 if (apmin >= tmp) 1177 return (EINVAL); 1178 #ifndef IPNOPRIVPORTS 1179 if (tmp < IPPORT_RESERVED) 1180 return (EINVAL); 1181 #endif /* IPNOPRIVPORTS */ 1182 break; 1183 1184 #ifndef IPNOPRIVPORTS 1185 case IPCTL_LOWPORTMIN: 1186 case IPV6CTL_LOWPORTMIN: 1187 if (tmp >= lpmax || 1188 tmp > IPPORT_RESERVEDMAX || 1189 tmp < IPPORT_RESERVEDMIN) 1190 return (EINVAL); 1191 break; 1192 1193 case IPCTL_LOWPORTMAX: 1194 case IPV6CTL_LOWPORTMAX: 1195 if (lpmin >= tmp || 1196 tmp > IPPORT_RESERVEDMAX || 1197 tmp < IPPORT_RESERVEDMIN) 1198 return (EINVAL); 1199 break; 1200 #endif /* IPNOPRIVPORTS */ 1201 1202 default: 1203 return (EINVAL); 1204 } 1205 1206 *(int*)rnode->sysctl_data = tmp; 1207 1208 return (0); 1209 } 1210 1211 static inline int 1212 copyout_uid(struct socket *sockp, void *oldp, size_t *oldlenp) 1213 { 1214 size_t sz; 1215 int error; 1216 uid_t uid; 1217 1218 uid = kauth_cred_geteuid(sockp->so_cred); 1219 if (oldp) { 1220 sz = MIN(sizeof(uid), *oldlenp); 1221 error = copyout(&uid, oldp, sz); 1222 if (error) 1223 return error; 1224 } 1225 *oldlenp = sizeof(uid); 1226 return 0; 1227 } 1228 1229 static inline int 1230 inet4_ident_core(struct in_addr raddr, u_int rport, 1231 struct in_addr laddr, u_int lport, 1232 void *oldp, size_t *oldlenp, 1233 struct lwp *l, int dodrop) 1234 { 1235 inpcb_t *inp; 1236 struct socket *so; 1237 1238 inp = inpcb_lookup(tcbtable, raddr, rport, laddr, lport, NULL); 1239 if (inp == NULL || (so = inpcb_get_socket(inp)) == NULL) 1240 return ESRCH; 1241 1242 if (dodrop) { 1243 struct tcpcb *tp; 1244 int error; 1245 1246 if ((tp = intotcpcb(inp)) == NULL || 1247 (so->so_options & SO_ACCEPTCONN) != 0) 1248 return ESRCH; 1249 1250 error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_SOCKET, 1251 KAUTH_REQ_NETWORK_SOCKET_DROP, so, tp, NULL); 1252 if (error) { 1253 return error; 1254 } 1255 1256 (void)tcp_drop(tp, ECONNABORTED); 1257 return 0; 1258 } 1259 1260 return copyout_uid(so, oldp, oldlenp); 1261 } 1262 1263 #ifdef INET6 1264 static inline int 1265 inet6_ident_core(struct in6_addr *raddr, u_int rport, 1266 struct in6_addr *laddr, u_int lport, 1267 void *oldp, size_t *oldlenp, 1268 struct lwp *l, int dodrop) 1269 { 1270 struct in6pcb *in6p; 1271 struct socket *sockp; 1272 1273 in6p = in6_pcblookup_connect(tcbtable, raddr, rport, laddr, lport, 0, 0); 1274 1275 if (in6p == NULL || (sockp = in6p->in6p_socket) == NULL) 1276 return ESRCH; 1277 1278 if (dodrop) { 1279 struct tcpcb *tp; 1280 int error; 1281 1282 if (in6p == NULL || (tp = in6totcpcb(in6p)) == NULL || 1283 (in6p->in6p_socket->so_options & SO_ACCEPTCONN) != 0) 1284 return ESRCH; 1285 1286 error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_SOCKET, 1287 KAUTH_REQ_NETWORK_SOCKET_DROP, in6p->in6p_socket, tp, NULL); 1288 if (error) 1289 return (error); 1290 1291 (void)tcp_drop(tp, ECONNABORTED); 1292 return 0; 1293 } 1294 else 1295 return copyout_uid(sockp, oldp, oldlenp); 1296 } 1297 #endif 1298 1299 /* 1300 * sysctl helper routine for the net.inet.tcp.drop and 1301 * net.inet6.tcp6.drop nodes. 1302 */ 1303 #define sysctl_net_inet_tcp_drop sysctl_net_inet_tcp_ident 1304 1305 /* 1306 * sysctl helper routine for the net.inet.tcp.ident and 1307 * net.inet6.tcp6.ident nodes. contains backwards compat code for the 1308 * old way of looking up the ident information for ipv4 which involves 1309 * stuffing the port/addr pairs into the mib lookup. 1310 */ 1311 static int 1312 sysctl_net_inet_tcp_ident(SYSCTLFN_ARGS) 1313 { 1314 #ifdef INET 1315 struct sockaddr_in *si4[2]; 1316 #endif /* INET */ 1317 #ifdef INET6 1318 struct sockaddr_in6 *si6[2]; 1319 #endif /* INET6 */ 1320 struct sockaddr_storage sa[2]; 1321 int error, pf, dodrop; 1322 1323 dodrop = name[-1] == TCPCTL_DROP; 1324 if (dodrop) { 1325 if (oldp != NULL || *oldlenp != 0) 1326 return EINVAL; 1327 if (newp == NULL) 1328 return EPERM; 1329 if (newlen < sizeof(sa)) 1330 return ENOMEM; 1331 } 1332 if (namelen != 4 && namelen != 0) 1333 return EINVAL; 1334 if (name[-2] != IPPROTO_TCP) 1335 return EINVAL; 1336 pf = name[-3]; 1337 1338 /* old style lookup, ipv4 only */ 1339 if (namelen == 4) { 1340 #ifdef INET 1341 struct in_addr laddr, raddr; 1342 u_int lport, rport; 1343 1344 if (pf != PF_INET) 1345 return EPROTONOSUPPORT; 1346 raddr.s_addr = (uint32_t)name[0]; 1347 rport = (u_int)name[1]; 1348 laddr.s_addr = (uint32_t)name[2]; 1349 lport = (u_int)name[3]; 1350 1351 mutex_enter(softnet_lock); 1352 error = inet4_ident_core(raddr, rport, laddr, lport, 1353 oldp, oldlenp, l, dodrop); 1354 mutex_exit(softnet_lock); 1355 return error; 1356 #else /* INET */ 1357 return EINVAL; 1358 #endif /* INET */ 1359 } 1360 1361 if (newp == NULL || newlen != sizeof(sa)) 1362 return EINVAL; 1363 error = copyin(newp, &sa, newlen); 1364 if (error) 1365 return error; 1366 1367 /* 1368 * requested families must match 1369 */ 1370 if (pf != sa[0].ss_family || sa[0].ss_family != sa[1].ss_family) 1371 return EINVAL; 1372 1373 switch (pf) { 1374 #ifdef INET6 1375 case PF_INET6: 1376 si6[0] = (struct sockaddr_in6*)&sa[0]; 1377 si6[1] = (struct sockaddr_in6*)&sa[1]; 1378 if (si6[0]->sin6_len != sizeof(*si6[0]) || 1379 si6[1]->sin6_len != sizeof(*si6[1])) 1380 return EINVAL; 1381 1382 if (!IN6_IS_ADDR_V4MAPPED(&si6[0]->sin6_addr) && 1383 !IN6_IS_ADDR_V4MAPPED(&si6[1]->sin6_addr)) { 1384 error = sa6_embedscope(si6[0], ip6_use_defzone); 1385 if (error) 1386 return error; 1387 error = sa6_embedscope(si6[1], ip6_use_defzone); 1388 if (error) 1389 return error; 1390 1391 mutex_enter(softnet_lock); 1392 error = inet6_ident_core(&si6[0]->sin6_addr, 1393 si6[0]->sin6_port, &si6[1]->sin6_addr, 1394 si6[1]->sin6_port, oldp, oldlenp, l, dodrop); 1395 mutex_exit(softnet_lock); 1396 return error; 1397 } 1398 1399 if (IN6_IS_ADDR_V4MAPPED(&si6[0]->sin6_addr) != 1400 IN6_IS_ADDR_V4MAPPED(&si6[1]->sin6_addr)) 1401 return EINVAL; 1402 1403 in6_sin6_2_sin_in_sock((struct sockaddr *)&sa[0]); 1404 in6_sin6_2_sin_in_sock((struct sockaddr *)&sa[1]); 1405 /*FALLTHROUGH*/ 1406 #endif /* INET6 */ 1407 #ifdef INET 1408 case PF_INET: 1409 si4[0] = (struct sockaddr_in*)&sa[0]; 1410 si4[1] = (struct sockaddr_in*)&sa[1]; 1411 if (si4[0]->sin_len != sizeof(*si4[0]) || 1412 si4[0]->sin_len != sizeof(*si4[1])) 1413 return EINVAL; 1414 1415 mutex_enter(softnet_lock); 1416 error = inet4_ident_core(si4[0]->sin_addr, si4[0]->sin_port, 1417 si4[1]->sin_addr, si4[1]->sin_port, 1418 oldp, oldlenp, l, dodrop); 1419 mutex_exit(softnet_lock); 1420 return error; 1421 #endif /* INET */ 1422 default: 1423 return EPROTONOSUPPORT; 1424 } 1425 } 1426 1427 static int 1428 sysctl_tcp_congctl(SYSCTLFN_ARGS) 1429 { 1430 struct sysctlnode node; 1431 int error; 1432 char newname[TCPCC_MAXLEN]; 1433 1434 strlcpy(newname, tcp_congctl_global_name, sizeof(newname) - 1); 1435 1436 node = *rnode; 1437 node.sysctl_data = newname; 1438 node.sysctl_size = sizeof(newname); 1439 1440 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1441 1442 if (error || 1443 newp == NULL || 1444 strncmp(newname, tcp_congctl_global_name, sizeof(newname)) == 0) 1445 return error; 1446 1447 mutex_enter(softnet_lock); 1448 error = tcp_congctl_select(NULL, newname); 1449 mutex_exit(softnet_lock); 1450 1451 return error; 1452 } 1453 1454 static int 1455 sysctl_tcp_init_win(SYSCTLFN_ARGS) 1456 { 1457 int error; 1458 u_int iw; 1459 struct sysctlnode node; 1460 1461 iw = *(u_int *)rnode->sysctl_data; 1462 node = *rnode; 1463 node.sysctl_data = &iw; 1464 node.sysctl_size = sizeof(iw); 1465 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1466 if (error || newp == NULL) 1467 return error; 1468 1469 if (iw >= __arraycount(tcp_init_win_max)) 1470 return EINVAL; 1471 *(u_int *)rnode->sysctl_data = iw; 1472 return 0; 1473 } 1474 1475 static int 1476 sysctl_tcp_keep(SYSCTLFN_ARGS) 1477 { 1478 int error; 1479 u_int tmp; 1480 struct sysctlnode node; 1481 1482 node = *rnode; 1483 tmp = *(u_int *)rnode->sysctl_data; 1484 node.sysctl_data = &tmp; 1485 1486 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1487 if (error || newp == NULL) 1488 return error; 1489 1490 mutex_enter(softnet_lock); 1491 1492 *(u_int *)rnode->sysctl_data = tmp; 1493 tcp_tcpcb_template(); /* update the template */ 1494 1495 mutex_exit(softnet_lock); 1496 return 0; 1497 } 1498 1499 static int 1500 sysctl_net_inet_tcp_stats(SYSCTLFN_ARGS) 1501 { 1502 1503 return (NETSTAT_SYSCTL(tcpstat_percpu, TCP_NSTATS)); 1504 } 1505 1506 /* 1507 * this (second stage) setup routine is a replacement for tcp_sysctl() 1508 * (which is currently used for ipv4 and ipv6) 1509 */ 1510 static void 1511 sysctl_net_inet_tcp_setup2(struct sysctllog **clog, int pf, const char *pfname, 1512 const char *tcpname) 1513 { 1514 const struct sysctlnode *sack_node; 1515 const struct sysctlnode *abc_node; 1516 const struct sysctlnode *ecn_node; 1517 const struct sysctlnode *congctl_node; 1518 const struct sysctlnode *mslt_node; 1519 const struct sysctlnode *vtw_node; 1520 #ifdef TCP_DEBUG 1521 extern struct tcp_debug tcp_debug[TCP_NDEBUG]; 1522 extern int tcp_debx; 1523 #endif 1524 1525 sysctl_createv(clog, 0, NULL, NULL, 1526 CTLFLAG_PERMANENT, 1527 CTLTYPE_NODE, "net", NULL, 1528 NULL, 0, NULL, 0, 1529 CTL_NET, CTL_EOL); 1530 sysctl_createv(clog, 0, NULL, NULL, 1531 CTLFLAG_PERMANENT, 1532 CTLTYPE_NODE, pfname, NULL, 1533 NULL, 0, NULL, 0, 1534 CTL_NET, pf, CTL_EOL); 1535 sysctl_createv(clog, 0, NULL, NULL, 1536 CTLFLAG_PERMANENT, 1537 CTLTYPE_NODE, tcpname, 1538 SYSCTL_DESCR("TCP related settings"), 1539 NULL, 0, NULL, 0, 1540 CTL_NET, pf, IPPROTO_TCP, CTL_EOL); 1541 1542 sysctl_createv(clog, 0, NULL, NULL, 1543 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1544 CTLTYPE_INT, "rfc1323", 1545 SYSCTL_DESCR("Enable RFC1323 TCP extensions"), 1546 NULL, 0, &tcp_do_rfc1323, 0, 1547 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RFC1323, CTL_EOL); 1548 sysctl_createv(clog, 0, NULL, NULL, 1549 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1550 CTLTYPE_INT, "sendspace", 1551 SYSCTL_DESCR("Default TCP send buffer size"), 1552 NULL, 0, &tcp_sendspace, 0, 1553 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SENDSPACE, CTL_EOL); 1554 sysctl_createv(clog, 0, NULL, NULL, 1555 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1556 CTLTYPE_INT, "recvspace", 1557 SYSCTL_DESCR("Default TCP receive buffer size"), 1558 NULL, 0, &tcp_recvspace, 0, 1559 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RECVSPACE, CTL_EOL); 1560 sysctl_createv(clog, 0, NULL, NULL, 1561 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1562 CTLTYPE_INT, "mssdflt", 1563 SYSCTL_DESCR("Default maximum segment size"), 1564 sysctl_net_inet_tcp_mssdflt, 0, &tcp_mssdflt, 0, 1565 CTL_NET, pf, IPPROTO_TCP, TCPCTL_MSSDFLT, CTL_EOL); 1566 sysctl_createv(clog, 0, NULL, NULL, 1567 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1568 CTLTYPE_INT, "minmss", 1569 SYSCTL_DESCR("Lower limit for TCP maximum segment size"), 1570 NULL, 0, &tcp_minmss, 0, 1571 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1572 sysctl_createv(clog, 0, NULL, NULL, 1573 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1574 CTLTYPE_INT, "msl", 1575 SYSCTL_DESCR("Maximum Segment Life"), 1576 NULL, 0, &tcp_msl, 0, 1577 CTL_NET, pf, IPPROTO_TCP, TCPCTL_MSL, CTL_EOL); 1578 sysctl_createv(clog, 0, NULL, NULL, 1579 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1580 CTLTYPE_INT, "syn_cache_limit", 1581 SYSCTL_DESCR("Maximum number of entries in the TCP " 1582 "compressed state engine"), 1583 NULL, 0, &tcp_syn_cache_limit, 0, 1584 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SYN_CACHE_LIMIT, 1585 CTL_EOL); 1586 sysctl_createv(clog, 0, NULL, NULL, 1587 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1588 CTLTYPE_INT, "syn_bucket_limit", 1589 SYSCTL_DESCR("Maximum number of entries per hash " 1590 "bucket in the TCP compressed state " 1591 "engine"), 1592 NULL, 0, &tcp_syn_bucket_limit, 0, 1593 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SYN_BUCKET_LIMIT, 1594 CTL_EOL); 1595 #if 0 /* obsoleted */ 1596 sysctl_createv(clog, 0, NULL, NULL, 1597 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1598 CTLTYPE_INT, "syn_cache_interval", 1599 SYSCTL_DESCR("TCP compressed state engine's timer interval"), 1600 NULL, 0, &tcp_syn_cache_interval, 0, 1601 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SYN_CACHE_INTER, 1602 CTL_EOL); 1603 #endif 1604 sysctl_createv(clog, 0, NULL, NULL, 1605 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1606 CTLTYPE_INT, "init_win", 1607 SYSCTL_DESCR("Initial TCP congestion window"), 1608 sysctl_tcp_init_win, 0, &tcp_init_win, 0, 1609 CTL_NET, pf, IPPROTO_TCP, TCPCTL_INIT_WIN, CTL_EOL); 1610 sysctl_createv(clog, 0, NULL, NULL, 1611 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1612 CTLTYPE_INT, "mss_ifmtu", 1613 SYSCTL_DESCR("Use interface MTU for calculating MSS"), 1614 NULL, 0, &tcp_mss_ifmtu, 0, 1615 CTL_NET, pf, IPPROTO_TCP, TCPCTL_MSS_IFMTU, CTL_EOL); 1616 sysctl_createv(clog, 0, NULL, &sack_node, 1617 CTLFLAG_PERMANENT, 1618 CTLTYPE_NODE, "sack", 1619 SYSCTL_DESCR("RFC2018 Selective ACKnowledgement tunables"), 1620 NULL, 0, NULL, 0, 1621 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_EOL); 1622 1623 /* Congctl subtree */ 1624 sysctl_createv(clog, 0, NULL, &congctl_node, 1625 CTLFLAG_PERMANENT, 1626 CTLTYPE_NODE, "congctl", 1627 SYSCTL_DESCR("TCP Congestion Control"), 1628 NULL, 0, NULL, 0, 1629 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1630 sysctl_createv(clog, 0, &congctl_node, NULL, 1631 CTLFLAG_PERMANENT, 1632 CTLTYPE_STRING, "available", 1633 SYSCTL_DESCR("Available Congestion Control Mechanisms"), 1634 NULL, 0, tcp_congctl_avail, 0, CTL_CREATE, CTL_EOL); 1635 sysctl_createv(clog, 0, &congctl_node, NULL, 1636 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1637 CTLTYPE_STRING, "selected", 1638 SYSCTL_DESCR("Selected Congestion Control Mechanism"), 1639 sysctl_tcp_congctl, 0, NULL, TCPCC_MAXLEN, 1640 CTL_CREATE, CTL_EOL); 1641 1642 sysctl_createv(clog, 0, NULL, NULL, 1643 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1644 CTLTYPE_INT, "win_scale", 1645 SYSCTL_DESCR("Use RFC1323 window scale options"), 1646 NULL, 0, &tcp_do_win_scale, 0, 1647 CTL_NET, pf, IPPROTO_TCP, TCPCTL_WSCALE, CTL_EOL); 1648 sysctl_createv(clog, 0, NULL, NULL, 1649 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1650 CTLTYPE_INT, "timestamps", 1651 SYSCTL_DESCR("Use RFC1323 time stamp options"), 1652 NULL, 0, &tcp_do_timestamps, 0, 1653 CTL_NET, pf, IPPROTO_TCP, TCPCTL_TSTAMP, CTL_EOL); 1654 sysctl_createv(clog, 0, NULL, NULL, 1655 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1656 CTLTYPE_INT, "compat_42", 1657 SYSCTL_DESCR("Enable workarounds for 4.2BSD TCP bugs"), 1658 NULL, 0, &tcp_compat_42, 0, 1659 CTL_NET, pf, IPPROTO_TCP, TCPCTL_COMPAT_42, CTL_EOL); 1660 sysctl_createv(clog, 0, NULL, NULL, 1661 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1662 CTLTYPE_INT, "cwm", 1663 SYSCTL_DESCR("Hughes/Touch/Heidemann Congestion Window " 1664 "Monitoring"), 1665 NULL, 0, &tcp_cwm, 0, 1666 CTL_NET, pf, IPPROTO_TCP, TCPCTL_CWM, CTL_EOL); 1667 sysctl_createv(clog, 0, NULL, NULL, 1668 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1669 CTLTYPE_INT, "cwm_burstsize", 1670 SYSCTL_DESCR("Congestion Window Monitoring allowed " 1671 "burst count in packets"), 1672 NULL, 0, &tcp_cwm_burstsize, 0, 1673 CTL_NET, pf, IPPROTO_TCP, TCPCTL_CWM_BURSTSIZE, 1674 CTL_EOL); 1675 sysctl_createv(clog, 0, NULL, NULL, 1676 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1677 CTLTYPE_INT, "ack_on_push", 1678 SYSCTL_DESCR("Immediately return ACK when PSH is " 1679 "received"), 1680 NULL, 0, &tcp_ack_on_push, 0, 1681 CTL_NET, pf, IPPROTO_TCP, TCPCTL_ACK_ON_PUSH, CTL_EOL); 1682 sysctl_createv(clog, 0, NULL, NULL, 1683 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1684 CTLTYPE_INT, "keepidle", 1685 SYSCTL_DESCR("Allowed connection idle ticks before a " 1686 "keepalive probe is sent"), 1687 sysctl_tcp_keep, 0, &tcp_keepidle, 0, 1688 CTL_NET, pf, IPPROTO_TCP, TCPCTL_KEEPIDLE, CTL_EOL); 1689 sysctl_createv(clog, 0, NULL, NULL, 1690 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1691 CTLTYPE_INT, "keepintvl", 1692 SYSCTL_DESCR("Ticks before next keepalive probe is sent"), 1693 sysctl_tcp_keep, 0, &tcp_keepintvl, 0, 1694 CTL_NET, pf, IPPROTO_TCP, TCPCTL_KEEPINTVL, CTL_EOL); 1695 sysctl_createv(clog, 0, NULL, NULL, 1696 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1697 CTLTYPE_INT, "keepcnt", 1698 SYSCTL_DESCR("Number of keepalive probes to send"), 1699 sysctl_tcp_keep, 0, &tcp_keepcnt, 0, 1700 CTL_NET, pf, IPPROTO_TCP, TCPCTL_KEEPCNT, CTL_EOL); 1701 sysctl_createv(clog, 0, NULL, NULL, 1702 CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE, 1703 CTLTYPE_INT, "slowhz", 1704 SYSCTL_DESCR("Keepalive ticks per second"), 1705 NULL, PR_SLOWHZ, NULL, 0, 1706 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SLOWHZ, CTL_EOL); 1707 sysctl_createv(clog, 0, NULL, NULL, 1708 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1709 CTLTYPE_INT, "log_refused", 1710 SYSCTL_DESCR("Log refused TCP connections"), 1711 NULL, 0, &tcp_log_refused, 0, 1712 CTL_NET, pf, IPPROTO_TCP, TCPCTL_LOG_REFUSED, CTL_EOL); 1713 #if 0 /* obsoleted */ 1714 sysctl_createv(clog, 0, NULL, NULL, 1715 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1716 CTLTYPE_INT, "rstratelimit", NULL, 1717 NULL, 0, &tcp_rst_ratelim, 0, 1718 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RSTRATELIMIT, CTL_EOL); 1719 #endif 1720 sysctl_createv(clog, 0, NULL, NULL, 1721 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1722 CTLTYPE_INT, "rstppslimit", 1723 SYSCTL_DESCR("Maximum number of RST packets to send " 1724 "per second"), 1725 NULL, 0, &tcp_rst_ppslim, 0, 1726 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RSTPPSLIMIT, CTL_EOL); 1727 sysctl_createv(clog, 0, NULL, NULL, 1728 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1729 CTLTYPE_INT, "delack_ticks", 1730 SYSCTL_DESCR("Number of ticks to delay sending an ACK"), 1731 NULL, 0, &tcp_delack_ticks, 0, 1732 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DELACK_TICKS, CTL_EOL); 1733 sysctl_createv(clog, 0, NULL, NULL, 1734 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1735 CTLTYPE_INT, "init_win_local", 1736 SYSCTL_DESCR("Initial TCP window size (in segments)"), 1737 sysctl_tcp_init_win, 0, &tcp_init_win_local, 0, 1738 CTL_NET, pf, IPPROTO_TCP, TCPCTL_INIT_WIN_LOCAL, 1739 CTL_EOL); 1740 sysctl_createv(clog, 0, NULL, NULL, 1741 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1742 CTLTYPE_STRUCT, "ident", 1743 SYSCTL_DESCR("RFC1413 Identification Protocol lookups"), 1744 sysctl_net_inet_tcp_ident, 0, NULL, sizeof(uid_t), 1745 CTL_NET, pf, IPPROTO_TCP, TCPCTL_IDENT, CTL_EOL); 1746 sysctl_createv(clog, 0, NULL, NULL, 1747 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1748 CTLTYPE_INT, "do_loopback_cksum", 1749 SYSCTL_DESCR("Perform TCP checksum on loopback"), 1750 NULL, 0, &tcp_do_loopback_cksum, 0, 1751 CTL_NET, pf, IPPROTO_TCP, TCPCTL_LOOPBACKCKSUM, 1752 CTL_EOL); 1753 sysctl_createv(clog, 0, NULL, NULL, 1754 CTLFLAG_PERMANENT, 1755 CTLTYPE_STRUCT, "pcblist", 1756 SYSCTL_DESCR("TCP protocol control block list"), 1757 sysctl_inpcblist, 0, tcbtable, 0, 1758 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, 1759 CTL_EOL); 1760 sysctl_createv(clog, 0, NULL, NULL, 1761 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1762 CTLTYPE_INT, "keepinit", 1763 SYSCTL_DESCR("Ticks before initial tcp connection times out"), 1764 sysctl_tcp_keep, 0, &tcp_keepinit, 0, 1765 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1766 1767 /* TCP socket buffers auto-sizing nodes */ 1768 sysctl_createv(clog, 0, NULL, NULL, 1769 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1770 CTLTYPE_INT, "recvbuf_auto", 1771 SYSCTL_DESCR("Enable automatic receive " 1772 "buffer sizing (experimental)"), 1773 NULL, 0, &tcp_do_autorcvbuf, 0, 1774 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1775 sysctl_createv(clog, 0, NULL, NULL, 1776 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1777 CTLTYPE_INT, "recvbuf_inc", 1778 SYSCTL_DESCR("Incrementor step size of " 1779 "automatic receive buffer"), 1780 NULL, 0, &tcp_autorcvbuf_inc, 0, 1781 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1782 sysctl_createv(clog, 0, NULL, NULL, 1783 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1784 CTLTYPE_INT, "recvbuf_max", 1785 SYSCTL_DESCR("Max size of automatic receive buffer"), 1786 NULL, 0, &tcp_autorcvbuf_max, 0, 1787 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1788 1789 sysctl_createv(clog, 0, NULL, NULL, 1790 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1791 CTLTYPE_INT, "sendbuf_auto", 1792 SYSCTL_DESCR("Enable automatic send " 1793 "buffer sizing (experimental)"), 1794 NULL, 0, &tcp_do_autosndbuf, 0, 1795 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1796 sysctl_createv(clog, 0, NULL, NULL, 1797 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1798 CTLTYPE_INT, "sendbuf_inc", 1799 SYSCTL_DESCR("Incrementor step size of " 1800 "automatic send buffer"), 1801 NULL, 0, &tcp_autosndbuf_inc, 0, 1802 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1803 sysctl_createv(clog, 0, NULL, NULL, 1804 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1805 CTLTYPE_INT, "sendbuf_max", 1806 SYSCTL_DESCR("Max size of automatic send buffer"), 1807 NULL, 0, &tcp_autosndbuf_max, 0, 1808 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1809 1810 /* ECN subtree */ 1811 sysctl_createv(clog, 0, NULL, &ecn_node, 1812 CTLFLAG_PERMANENT, 1813 CTLTYPE_NODE, "ecn", 1814 SYSCTL_DESCR("RFC3168 Explicit Congestion Notification"), 1815 NULL, 0, NULL, 0, 1816 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1817 sysctl_createv(clog, 0, &ecn_node, NULL, 1818 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1819 CTLTYPE_INT, "enable", 1820 SYSCTL_DESCR("Enable TCP Explicit Congestion " 1821 "Notification"), 1822 NULL, 0, &tcp_do_ecn, 0, CTL_CREATE, CTL_EOL); 1823 sysctl_createv(clog, 0, &ecn_node, NULL, 1824 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1825 CTLTYPE_INT, "maxretries", 1826 SYSCTL_DESCR("Number of times to retry ECN setup " 1827 "before disabling ECN on the connection"), 1828 NULL, 0, &tcp_ecn_maxretries, 0, CTL_CREATE, CTL_EOL); 1829 1830 /* SACK gets it's own little subtree. */ 1831 sysctl_createv(clog, 0, NULL, &sack_node, 1832 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1833 CTLTYPE_INT, "enable", 1834 SYSCTL_DESCR("Enable RFC2018 Selective ACKnowledgement"), 1835 NULL, 0, &tcp_do_sack, 0, 1836 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 1837 sysctl_createv(clog, 0, NULL, &sack_node, 1838 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1839 CTLTYPE_INT, "maxholes", 1840 SYSCTL_DESCR("Maximum number of TCP SACK holes allowed per connection"), 1841 NULL, 0, &tcp_sack_tp_maxholes, 0, 1842 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 1843 sysctl_createv(clog, 0, NULL, &sack_node, 1844 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1845 CTLTYPE_INT, "globalmaxholes", 1846 SYSCTL_DESCR("Global maximum number of TCP SACK holes"), 1847 NULL, 0, &tcp_sack_globalmaxholes, 0, 1848 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 1849 sysctl_createv(clog, 0, NULL, &sack_node, 1850 CTLFLAG_PERMANENT, 1851 CTLTYPE_INT, "globalholes", 1852 SYSCTL_DESCR("Global number of TCP SACK holes"), 1853 NULL, 0, &tcp_sack_globalholes, 0, 1854 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 1855 1856 sysctl_createv(clog, 0, NULL, NULL, 1857 CTLFLAG_PERMANENT, 1858 CTLTYPE_STRUCT, "stats", 1859 SYSCTL_DESCR("TCP statistics"), 1860 sysctl_net_inet_tcp_stats, 0, NULL, 0, 1861 CTL_NET, pf, IPPROTO_TCP, TCPCTL_STATS, 1862 CTL_EOL); 1863 sysctl_createv(clog, 0, NULL, NULL, 1864 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1865 CTLTYPE_INT, "local_by_rtt", 1866 SYSCTL_DESCR("Use RTT estimator to decide which hosts " 1867 "are local"), 1868 NULL, 0, &tcp_rttlocal, 0, 1869 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1870 #ifdef TCP_DEBUG 1871 sysctl_createv(clog, 0, NULL, NULL, 1872 CTLFLAG_PERMANENT, 1873 CTLTYPE_STRUCT, "debug", 1874 SYSCTL_DESCR("TCP sockets debug information"), 1875 NULL, 0, &tcp_debug, sizeof(tcp_debug), 1876 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DEBUG, 1877 CTL_EOL); 1878 sysctl_createv(clog, 0, NULL, NULL, 1879 CTLFLAG_PERMANENT, 1880 CTLTYPE_INT, "debx", 1881 SYSCTL_DESCR("Number of TCP debug sockets messages"), 1882 NULL, 0, &tcp_debx, sizeof(tcp_debx), 1883 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DEBX, 1884 CTL_EOL); 1885 #endif 1886 sysctl_createv(clog, 0, NULL, NULL, 1887 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1888 CTLTYPE_STRUCT, "drop", 1889 SYSCTL_DESCR("TCP drop connection"), 1890 sysctl_net_inet_tcp_drop, 0, NULL, 0, 1891 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DROP, CTL_EOL); 1892 sysctl_createv(clog, 0, NULL, NULL, 1893 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1894 CTLTYPE_INT, "iss_hash", 1895 SYSCTL_DESCR("Enable RFC 1948 ISS by cryptographic " 1896 "hash computation"), 1897 NULL, 0, &tcp_do_rfc1948, sizeof(tcp_do_rfc1948), 1898 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, 1899 CTL_EOL); 1900 1901 /* ABC subtree */ 1902 1903 sysctl_createv(clog, 0, NULL, &abc_node, 1904 CTLFLAG_PERMANENT, CTLTYPE_NODE, "abc", 1905 SYSCTL_DESCR("RFC3465 Appropriate Byte Counting (ABC)"), 1906 NULL, 0, NULL, 0, 1907 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1908 sysctl_createv(clog, 0, &abc_node, NULL, 1909 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1910 CTLTYPE_INT, "enable", 1911 SYSCTL_DESCR("Enable RFC3465 Appropriate Byte Counting"), 1912 NULL, 0, &tcp_do_abc, 0, CTL_CREATE, CTL_EOL); 1913 sysctl_createv(clog, 0, &abc_node, NULL, 1914 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1915 CTLTYPE_INT, "aggressive", 1916 SYSCTL_DESCR("1: L=2*SMSS 0: L=1*SMSS"), 1917 NULL, 0, &tcp_abc_aggressive, 0, CTL_CREATE, CTL_EOL); 1918 1919 /* MSL tuning subtree */ 1920 1921 sysctl_createv(clog, 0, NULL, &mslt_node, 1922 CTLFLAG_PERMANENT, CTLTYPE_NODE, "mslt", 1923 SYSCTL_DESCR("MSL Tuning for TIME_WAIT truncation"), 1924 NULL, 0, NULL, 0, 1925 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1926 sysctl_createv(clog, 0, &mslt_node, NULL, 1927 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1928 CTLTYPE_INT, "enable", 1929 SYSCTL_DESCR("Enable TIME_WAIT truncation"), 1930 NULL, 0, &tcp_msl_enable, 0, CTL_CREATE, CTL_EOL); 1931 sysctl_createv(clog, 0, &mslt_node, NULL, 1932 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1933 CTLTYPE_INT, "loopback", 1934 SYSCTL_DESCR("MSL value to use for loopback connections"), 1935 NULL, 0, &tcp_msl_loop, 0, CTL_CREATE, CTL_EOL); 1936 sysctl_createv(clog, 0, &mslt_node, NULL, 1937 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1938 CTLTYPE_INT, "local", 1939 SYSCTL_DESCR("MSL value to use for local connections"), 1940 NULL, 0, &tcp_msl_local, 0, CTL_CREATE, CTL_EOL); 1941 sysctl_createv(clog, 0, &mslt_node, NULL, 1942 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1943 CTLTYPE_INT, "remote", 1944 SYSCTL_DESCR("MSL value to use for remote connections"), 1945 NULL, 0, &tcp_msl_remote, 0, CTL_CREATE, CTL_EOL); 1946 sysctl_createv(clog, 0, &mslt_node, NULL, 1947 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1948 CTLTYPE_INT, "remote_threshold", 1949 SYSCTL_DESCR("RTT estimate value to promote local to remote"), 1950 NULL, 0, &tcp_msl_remote_threshold, 0, CTL_CREATE, CTL_EOL); 1951 1952 /* vestigial TIME_WAIT tuning subtree */ 1953 1954 sysctl_createv(clog, 0, NULL, &vtw_node, 1955 CTLFLAG_PERMANENT, CTLTYPE_NODE, "vtw", 1956 SYSCTL_DESCR("Tuning for Vestigial TIME_WAIT"), 1957 NULL, 0, NULL, 0, 1958 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1959 sysctl_createv(clog, 0, &vtw_node, NULL, 1960 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1961 CTLTYPE_INT, "enable", 1962 SYSCTL_DESCR("Enable Vestigial TIME_WAIT"), 1963 sysctl_tcp_vtw_enable, 0, 1964 (pf == AF_INET) ? &tcp4_vtw_enable : &tcp6_vtw_enable, 1965 0, CTL_CREATE, CTL_EOL); 1966 sysctl_createv(clog, 0, &vtw_node, NULL, 1967 CTLFLAG_PERMANENT|CTLFLAG_READONLY, 1968 CTLTYPE_INT, "entries", 1969 SYSCTL_DESCR("Maximum number of vestigial TIME_WAIT entries"), 1970 NULL, 0, &tcp_vtw_entries, 0, CTL_CREATE, CTL_EOL); 1971 } 1972 1973 void 1974 tcp_usrreq_init(void) 1975 { 1976 1977 #ifdef INET 1978 sysctl_net_inet_tcp_setup2(NULL, PF_INET, "inet", "tcp"); 1979 #endif 1980 #ifdef INET6 1981 sysctl_net_inet_tcp_setup2(NULL, PF_INET6, "inet6", "tcp6"); 1982 #endif 1983 } 1984 1985 PR_WRAP_USRREQ(tcp_usrreq) 1986 1987 #define tcp_usrreq tcp_usrreq_wrapper 1988 1989 const struct pr_usrreqs tcp_usrreqs = { 1990 .pr_attach = tcp_attach, 1991 .pr_detach = tcp_detach, 1992 .pr_generic = tcp_usrreq, 1993 }; 1994
Indexes created Sun Sep 21 20:09:37 GMT 2025